(C) PLOS One [1]. This unaltered content originally appeared in journals.plosone.org.
Licensed under Creative Commons Attribution (CC BY) license.
url:https://journals.plos.org/plosone/s/licenses-and-copyright

------------



Comparative molecular genomic analyses of a spontaneous rhesus macaque model of mismatch repair-deficient colorectal cancer

['Nejla Ozirmak Lermi', 'Department Of Clinical Cancer Prevention', 'The University Of Texas Md Anderson Cancer Center', 'Houston', 'Texas', 'United States Of America', 'School Of Health Professions', 'Stanton B. Gray', 'Michale E. Keeling Center For Comparative Medicine', 'Research']

Date: 2022-06

Colorectal cancer (CRC) remains the third most common cancer in the US with 15% of cases displaying Microsatellite Instability (MSI) secondary to Lynch Syndrome (LS) or somatic hypermethylation of the MLH1 promoter. A cohort of rhesus macaques from our institution developed spontaneous mismatch repair deficient (MMRd) CRC with a notable fraction harboring a pathogenic germline mutation in MLH1 (c.1029C<G, p.Tyr343Ter). Our study aimed to provide a detailed molecular characterization of rhesus CRC for cross-comparison with human MMRd CRC. We performed PCR-based MSI testing (n = 41), transcriptomics analysis (n = 35), reduced-representation bisulfite sequencing (RRBS) (n = 28), and MLH1 DNA methylation (n = 10) using next-generation sequencing (NGS) of rhesus CRC. Systems biology tools were used to perform gene set enrichment analysis (GSEA) for pathway discovery, consensus molecular subtyping (CMS), and somatic mutation profiling. Overall, the majority of rhesus tumors displayed high levels of MSI (MSI-H) and differential gene expression profiles that were consistent with known deregulated pathways in human CRC. DNA methylation analysis exposed differentially methylated patterns among MSI-H, MSI-L (MSI-low)/MSS (MS-stable) and LS tumors with MLH1 predominantly inactivated among sporadic MSI-H CRCs. The findings from this study support the use of rhesus macaques as an alternative animal model to mice to study carcinogenesis, develop immunotherapies and vaccines, and implement chemoprevention approaches relevant to sporadic MSI-H and LS CRC in humans.

CRC remains the third most common cancer diagnosed in the United States. Some CRC may arise spontaneously without any known risk factors, while others may arise in patients with strong family history associated to inherited genetic syndromes. Our study focused on a genetic condition known as Lynch Syndrome (LS), which significantly increases the risk of developing CRC as well as several different types of cancers. Biological tools and laboratory animal models available for studying hereditary CRC remain limited and not always directly translatable to human disease. Therefore, our study presents a comprehensive analysis of a spontaneous non-human primate (NHP) model used to study the genetic contribution in LS CRC. We performed a cross-comparison of different types of CRC in humans with tumors developed in monkeys to determine the accuracy of using this NHP model for studying early cancer development, treatment options, and prevention approaches in both hereditary and sporadic colorectal cancer displaying MMR deficiency.

Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: Dr. Vilar has a consulting or advisory role with Janssen Research and Development and Recursion Pharma. He has received research support from Janssen Research and Development. Please, note that these financial relations are not connected to the research reported in this manuscript.

Funding: This work was supported by a gift from the Feinberg Family Foundation and MDACC Institutional Research Grant (IRG) Program to E.V.; MD Anderson Internal Grant Award from Cattlemen for Cancer Research to S.G.; R24 OD011173 (US National Institutes of Health) to J.R.; and CA016672 (US National Institutes of Health/National Cancer Institute) to The University of Texas MD Anderson Cancer Center Core Support Grant. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Data Availability: The project data has been deposited in GEO. The data sets generated and analyzed during the current study can be accessed for re-analysis using the following link through GEO Series accession number GSE178383. ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE178383 ). To access total RNA-seq data, use the following GEO sub-series accession number GSE178381. ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE178381 ). To access RRBS data, use the following GEO sub-series accession number GSE178377. ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE178377 ).

This study characterized the genomic features of colorectal tumors in the KCCMR rhesus cohort using microsatellite marker testing, whole transcriptomics, and epigenomics coupled with systems biology tools, as illustrated in Fig 1 . Additionally, we cross-compared the current subtypes of CRC in humans with the rhesus model to evaluate the utility of rhesus for studying cancer development, and developing treatment modalities, and prevention approaches in sporadic MSI-H CRC and LS.

A cohort of specific pathogen free (SPF), Indian-origin rhesus macaques bred at The University of Texas MD Anderson Cancer Center (MDACC) Michale E. Keeling Center for Comparative Medicine and Research (KCCMR) spontaneously develops MSI/MMRd CRC, including a subset of animals harboring a pathogenic germline mutation in MLH1 (c.1029C<G, p.Tyr343Ter). This spontaneous mutation manifests into clinical and pathological features analogous to human LS, which suggests that these rhesus macaques may be an informative model organism for studying the biology of MMRd CRC [ 10 , 12 ].

Given the anatomic and physiologic similarities and the genomic homology between non-human primates (NHPs) and humans, researchers have used several species to develop therapies and vaccines to treat and eradicate human disease [ 4 , 5 ]. The rhesus macaque (Macaca mulatta), which shares 97.5% DNA sequence identity with humans in exons of protein-coding genes as well as close similarity in patterns of gene expression, has been an invaluable animal model for studying human pathophysiology [ 6 , 7 ]. Studies have shown that rhesus launch parallel immune responses to humans, thus making them another available animal model for ascertaining clinical translation of basic and pre-clinical findings in conjunction with other model organisms [ 8 – 11 ].

Presently, in vitro and ex vivo models, such as cell lines and organoids respectively, are commonly used to study CRC; however, the intrinsic nature of these models lack cellular heterogeneity and fail to recapitulate the tumor microenvironment (TME) observed in-vivo [ 2 ]. To tackle the limitations of in-vitro/ex-vivo cultures, mouse models (Mus musculus) have been leveraged to study CRC prevention, initiation, and progression. Although murine models of genetic inactivation of MMR genes exist, these models present critical differences to the human LS (MMRd) phenotype. For example, murine models with constitutional homozygous MMR gene inactivation have high rates of lymphoma formation, limiting the efficacy of these models. In an effort to circumvent this challenge, investigators have employed tissue-specific Cre recombinase-based inactivation of MMR genes; however, these mice predominantly develop tumors in the small intestine (as opposed to the large intestine in humans) [ 3 ]. Although all models bear imperfections, the limitations of cellular cultures and murine models warrant the need for novel model systems that can contribute to improve the clinical outcomes for both LS and MSI sporadic patients.

A better understanding of colorectal neoplasia arising in the setting of MSI/MMRd is urgently needed to tailor the use of early detection, prevention, and treatment interventions in this subset of CRC, including immunotherapies (e.g. checkpoint inhibitors) and the development of novel immuno-preventive regimens (e.g. vaccines). Such interventions are particularly needed for those with Lynch syndrome (LS), as they are at the highest risk of CRC as well as a range of other cancers. Unfortunately, no concrete model with higher translational value exists to study the nuanced carcinogenesis of MMRd CRC, which is a critical barrier for studying this subset of CRC and, consequently, to making advances in its detection, prevention, and treatment.

Colorectal cancer (CRC) remains the third leading cause of cancer-related deaths affecting both men and women [ 1 ]. Approximately 15% of CRC cases display microsatellite instability (MSI) secondary to a defective mismatch repair (MMRd) system that is recognized as a major carcinogenic pathway for CRC development. MMRd arises from either (1) an inherited germline mutation in one of four genes (MLH1, MSH2, MSH6 and PMS2) constituting the DNA MMR system followed by an acquired second hit in the wild-type allele of the same gene in colonic mucosa cells (i.e., Lynch syndrome), or (2) somatic inactivation of the MLH1 gene (i.e., MSI sporadic CRC).

We examined somatic variants of rhesus CRC from the total RNAseq data. Our data indicated that mutational rate in coding regions of rhesus CRC was similar in all tested samples ( S11A Fig ). Substitutions of cytosine to thymine were the most abundant somatic variants in rhesus CRC ( S11C Fig ). Commonly altered genes in human CRC were also mutated in rhesus such as APC, ARID1A, TGBRII, TP53, CTNNB1, PIK3CA, KRAS ( S11B and S12 Figs ). Due to the close relation found in humans between MSI-H status and BRAF mutations, we performed Sanger sequencing to assess the mutational status of the BRAF mutation hotspot V600E in rhesus CRC. While we did not detect BRAF V600E mutations among rhesus tumors, we did observe several types of BRAF somatic variants including missense, nonsense, in-frame, and frameshift deletions ( S13 Fig ).

We assigned a consensus molecular subtype (CMS) status to each tumor sample based on the nearest CMS probability ( S3 Table ). Overall, 52% (n = 10) of tumors were classified as CMS2, which corresponds to the canonical pathways of colorectal carcinogenesis; 21% (n = 4) were CMS1, which progresses through MSI and immune pathways; and 21% (n = 4) were CMS4, which develops through mesenchymal pathways. Only one tumor displayed mixed features (CMS1-CMS2) of a transition phenotype ( Fig 5D ).

(A-C) Significant gene expression pathways relevant to CRC biology are highlighted in (A) MSI-H (sporadic and LS) and (B) in MSI-L/MSS (sporadic and LS) compared to normal tissue samples. (C) Highlighted pathways are up regulated in MSI-H (sporadic and LS) rhesus CRC compared to MSI-L/MSS (sporadic and LS) rhesus CRC. BH-adjusted P-value≤0.05 was set as threshold for analysis; (D) CMS classification of rhesus CRC. The outer ring of circos plot represents CMS subtypes present in rhesus CRC with 52% of samples (n = 10) classifying as CMS2. Middle ring represents MSI status of samples, and inner ring indicates clinical categories of samples.

Gene set enrichment analysis (GSEA) was performed to discover relevant pathways in colorectal carcinogenesis using the ESTIMATE algorithm, which assesses immune and stromal cell admixtures in tumors, canonical, immune, and metabolic pathways ( Fig 5A–5C ) [ 15 , 16 ]. When compared with normal tissue samples, top pathways enriched in MSI-H tumors were involved in cell cycle regulation, crypt base dynamics, and integrin signaling. Conversely, metabolic pathways in MSI-H samples were downregulated compared to normal tissue ( Fig 5A ). A similar trend was observed for MSS/MSI-L tumor samples compared to normal ( Fig 5B ). Lastly, comparing the significant pathways between MSS/MSI-L and MSI-H, we observed an upregulation of key pathways involved in cell cycle regulation and MYC targeting in the MSI-H group ( Fig 5C ).

We then determined significantly differentially expressed genes (DEGs) between rhesus normal and tumor using a Benjamini-Hochberg (BH)-adjusted P-value≤0.05 and log2 fold change±1. We annotated genes using human orthologs ( S10B Fig ). Unsupervised hierarchical clustering using DEGs demonstrated that rhesus tumor tissue samples clustered separately from normal tissue samples, and rhesus MSS/MSI-L CRC were separated from MSI-H CRC samples. Notably, the MSI-L tumor sample from the rhesus LS animal (RM17) clustered with the MSI-H and LS group ( Fig 4C ). Using total RNAseq transcripts, we sought to validate the expression of MMR genes using the read counts from tumors and matched normal samples. MLH1 read counts in MSI-H CRC samples were significantly decreased compared to normal tissue samples (P-value<0.0001). As expected, animal RM02 with a MSS tumor showed more MLH1 read counts in tumor than matched normal. MSH6 gene read counts in MSI-H CRC samples were significantly more abundant than matched-normal samples (P-value<0.001). Differences of MSH2 and PMS2 gene read counts between the rhesus tumor and normal tissue samples were not significant ( S10C Fig ).

( A ) 3D principal component analysis (PCA) of rhesus CRC gene expression profiles show clear separation among sporadic MSI-H samples (green pyramids), sporadic MSS and MSI-L (purple spheres), Lynch syndrome (blue cubes), and normal tissue (red diamonds). Normal tissue samples clustered separately from tumor tissue samples; (B) Pearson’s correlation coefficient of mean expression levels across 101 significant genes from COAD-READ MSI-H tumor samples, COADREAD MSS tumor samples, COADREAD normal tissue samples, rhesus LS tumor samples, and rhesus normal tissue samples; (C) Significant differentially expressed genes (DEGs) between tumor and normal tissue samples. DEGs were found based on BH-adjusted P-value≤0.05 between rhesus colorectal normal and tumor. Pearson’s correlation was used to perform hierarchical clustering between rhesus tumor and normal tissue samples. Columns represent samples, and rows represent statistically significant differentially expressed genes. MSI bar displays MSI status of samples based on PCR-based MSI testing. Gray color represents normal, pink MSI-H, and magenta MSS and MSI-L tissue samples. MSI type bar displays MLH1 genotyping data with gray color representing normal, orange LS, purple sporadic MSI-L and MSS and red sporadic MSI-H tissue samples.

We performed whole transcriptome sequencing in 19 colorectal tumors and 16 matched normal mucosa samples with an average tumor purity estimates in silico of 66% ( S9 Fig ). Unsupervised 3D principal component analysis (PCA) of RNAseq data showed a clear separation between tumor and normal samples. However, samples from the rhesus LS, sporadic MSI-H, and MSS/MSI-L clustered together ( Figs 4A and S10A ). To further characterize the rhesus LS as a model of human MSI-H CRC, we compared rhesus LS tumor to human MSI-H (n = 96) and MSS (n = 440) CRC cases from The Cancer Genome Atlas (TCGA) colon and rectal adenocarcinoma projects (COAD and READ, respectively) using the edgeR package. A total of 101 orthologous genes demonstrated statistically significant changes (BH-adjusted P-value < 0.05) in the expression level by at least two-fold difference (log2FC≥1). Then, we aimed to compare global gene expression patterns among rhesus Lynch tumor samples (n = 21) and COADREAD MSI-H and MSS samples to check for their correlation, while using COADREAD (human, n = 54) and rhesus normal samples (n = 20) to control the distance between both species. Rhesus Lynch tumor samples correlated better with COADREAD MSI-H tumor samples (0.82) than that with COADREAD MSS samples (0.68) and normal samples (0.64, Fig 4B ), thus suggesting that our analysis had sufficient resolution to analyze tumor tissue similarities.

Lastly, we performed a dedicated methylation analysis of the MLH1 promoter (n = 10) using a methyl NGS panel ( S3 Table ). The location of the CpG regions from the transcription start site of the MLH1 gene were identified observing thirteen CpGs significantly methylated in rhesus sporadic MSI-H tumor samples compared to adjacent normal mucosa (P-value<0.05). The majority of the methylated CpG regions were within exon 1. There were no significant methylation differences between other tumor sub-groups (MSS/MSI-L) and normal tissue samples ( S8 Fig ) ; however, there was a clear trend of higher levels of MLH1 promoter methylation among rhesus sporadic MSI-H compared to MSS tumors as well as a notorious absence of MLH1 methylation in the only LS tumor tested, which is consistent with human CRC biology.

(A) 3D PCA of DNA methylation in rhesus specimens characterizing the trends exhibited by differentially methylated region profiles of sporadic MSI-H (green pyramid), sporadic MSS and MSI-L (purple cube), Lynch syndrome (blue sphere), and normal tissue (red diamond) samples. Each shape represents a tissue sample type. Each group clusters separately; however, sporadic MSS and MSI-L CRC samples are closer to normal tissue samples; (B) Hierarchical clustering of DNA methylation profiles assesses by CpG methylation using Pearson’s correlation. Distance displays the relationship between rhesus tumors and matched normal tissue samples with parameters set as distance method: “correlation”, clustering method: “ward”; (C) Significant differentially methylated regions (DMRs) of rhesus tumors displaying MSI-H and MSI-L/MSS phenotypes at FDR of 5%. PK1B, B4GALT7, GPR35, MYT1L and CYB5D2 are among hyper-methylated genes, and GRB10, SOD1, TMSB10 and CD52 are hypo-methylated.

As seen in human MSI CRC, the MSI testing and the transcriptomic profiling of rhesus MSI-H CRC suggested that the vast majority of rhesus CRC could be secondary to an epigenetic event. To determine the epigenetic contribution to rhesus CRC carcinogenesis, we analyzed global DNA methylation patterns in tumor (n = 14) and matched adjacent normal samples ( S3 Table ). Unsupervised 3D principal component analysis (PCA) of reduced-representation bisulfite sequencing (RRBS) data revealed clear clustering of sporadic MSI-H, MSI-L/MSS as well as normal mucosa with MSI-L/MSS tumors clustering closer to normal mucosa. LS rhesus tumors clustered according to their MSI status with three MSI-H closer to their sporadic counterparts and the MSI-L doing the same with the sporadic MSI-L/MSS group ( Fig 3A ). When we performed hierarchical clustering of DNA methylation profiles, a clear separation between sporadic rhesus MSI-H and MSI-L/MSS tumors became evident with normal colorectal samples closer to MSI-L/MSS tumors. The analysis confirmed the robustness of the three main clusters with unbiased P-value that were statistically for all the groups ( S6 Fig ) using two methods. Of note, rhesus LS tumors displaying MSI-H patterns (RM25, RM31) clustered together with sporadic MSI-H tumors and the one MSI-L rhesus LS (RM17) was closer to normal and rhesus MSS/MSI-L samples ( Fig 3B ). One rhesus LS animal with MSI-H phenotype (RM38) clustered together with normal and rhesus MSS/MSI-L samples, but the clustering branch of this animal was physically and statistically closer to MSI-H tumors than to MSI-L/MSS and normal tissues ( Figs 3B and S6 ). A total of 628 hypermethylated and 592 hypomethylated genes were identified as significant differentially methylated regions (DMRs) using a cut-off of FDR of 5% between the rhesus MSI-H (grouping both sporadic and LS) and MSI-L/MSS (sporadic and one LS tumor) involving some of following genes: PK1B, B4GALT7, GPR35, MYT1L and CYB5D2 (hypermethylated), and GRB10, SOD1, TMSB10 and CD52 (hypomethylated, Fig 3C ). In addition, we also detected a number of DMRs between rhesus tumor and adjacent normal colorectal mucosa at FDR of 5% ( S7A Fig ). A correlation analysis revealed a negative relation between DNA methylation and gene expression levels that showed a trend towards statistical significance (P-value = 0.1336, S7B Fig ), thus demonstrating that DNA methylation in rhesus CRC affected gene expression levels.

Using the newly designed rhesus MSI panel, we performed MSI testing in all tumors from the KCCMR cohort using matched normal samples as genomic reference (n = 41). c-kitRheBAT25, RheBAT26, and RheD18S58 markers were the most sensitive ( Fig 2E ). We validated the calls made in RheBAT26 and RheD18S58 using an alternative technique based on fragment analysis ( S5 Fig ). We classified rhesus tumors into the three classical categories (i.e. MSI-H, MSI-L, and MSS) by counting the number of unstable markers in each tumor, thus following the classical NCI recommendations [ 14 ]. Thirty-one samples were MSI-H (76%), six were MSI-L (15%), and four were MSS (10%, Fig 2F ). Two rhesus LS animals (RM11 and RM17) displayed an MSI-L phenotype and six rhesus LS animals (RM09, RM22, RM25, RM31, RM38 and RM41) presented an MSI-H phenotype ( Fig 2C and 2F ).

We developed a PCR-based MSI testing panel for rhesus CRC including orthologs of the most frequently used microsatellite markers in human CRC: BAT25, BAT26, BAT40, D10S197, D18S58, D2S123, D17S250, D5S346, β-catenin, and TGFβRII. Rhesus orthologs of D2S123, D17S250, and D5S346 markers did not contain adequate nucleotide repeats suitable to assess the presence of MSI. Hence, we excluded these markers from our rhesus MSI testing panel. Furthermore, the rhesus ortholog of BAT25 was not sensitive enough to determine MSI due to the interruption of the microsatellite tract by one nucleotide. Therefore, we substituted it with a novel MSI marker (named c-kitRheBAT25), which was identified by screening the whole sequence of c-kit to identify an uninterrupted repeat region. Overall, the rhesus CRC MSI testing panel included 6 markers: 4 mononucleotides (c-kitRheBAT25, RheBAT26, RheBAT40, RheTGFβRII) and 2 dinucleotides (RheD18S58, RheD10S197) ( S1 Table ). This panel offers an assessment of the functionality of the MMR system in these rhesus macaques.

We obtained IHC data from a total of 37 rhesus CRCs (n = 37) with 36 samples (97%) displaying loss of protein expression in MLH1 and/or PMS2. Only one animal (~3%) retained the expression of the MLH1-PMS2 heterodimer. This same animal also displayed complete stability of all the MSI markers, thus being MSS, and therefore considered as MMR proficient. Subsequently, we used this animal as a control for all further genomic analyses ( Figs 2D and S4 ).

We detected the presence of a previously described heterozygous germline stop codon mutation in exon 11 of MLH1 (c.1029C>G; p.Tyr343Ter, Figs 2C and S2 ) in 8 animals (~20%) [ 10 ], thus confirming the presence of a causative pathogenic mutation previously described in humans (herein these animals are referred to as rhesus Lynch) [ 12 ].The remaining 33 animals (80%) had the wild-type germline sequence of MLH1 (herein referred to as sporadic rhesus, Fig 2C ). The pedigree of rhesus Lynch animals revealed the autosomal dominant inheritance pattern of the MLH1 mutation and an inheritance pattern concordant with the Amsterdam criteria (the 3-2-1 rule) originally described in human LS ( S3 Fig ) [ 13 ].

(A) Animal ages at the time of diagnosis of CRC and subsequent euthanasia. The average age at death for the rhesus CRC cohort was 19.14 years. Red dots indicate the age of animals with MLH1 germline mutation; (B) Gender of rhesus cohort. The majority of animals in this cohort were female; (C) Lynch syndrome MLH1 germline mutation status. A total of eight (20%) carried a heterozygous MLH1 nonsense mutation (c.1029, C>G); (D) IHC assessment of rhesus CRC. The majority of rhesus tumor samples displayed loss of MLH1 and PMS2; (E) MSI testing of rhesus tumors. Newly designed MSI testing panel for rhesus CRC included six markers (RheBAT25, RheBAT26, RheBAT40, RheD10S197, RheD18S58, and RheTGFβRII) that were orthologs of commonly tested MSI loci in human tumors (BAT25, BAT26, BAT40, D10S197, D18S58, and TGF βRII). Overall, RheBAT25, RheBAT26, and RheD18S58 MSI markers were the most mutated MSI markers in rhesus CRC; (F) Summary of MSI status of rhesus tumors. Rhesus CRC were predom-inantly MSI-H (75%), and only six tumors (15%) were MSI-L, and four (10%) MSS.

We identified a total of 41 animals diagnosed with CRC at the time of necropsy. Specimens were collected between 2008 and 2019. All tumors were located in the right side of the colon (20 in the ascending colon, 16 in the ileocecal valve, and 4 in the cecum) with the exception of one jejunal tumor. The mean age at death was 19.1 years (range: 9 to 27, Fig 2A ) and 80% of animals were female ( Fig 2B ), consistent with overall population demographics of the cohort from which the animals were drawn. The average age at death was younger among LS animals compared to sporadic MSI-H macaques, but the difference did not reach statistical significant (17.75 vs 19.48 years, P-value = 0.2, S1 Fig ).

Discussion

Although cell cultures, organoids, and murine animal models are the most frequently used models in CRC research, these systems sometimes fail to recapitulate all the phenotypic features of MMRd CRC, thus limiting clinical translation to humans. To overcome some of the differences between humans and research models, investigators are also pursuing the use of alternative non-murine models such as dogs, cats, pigs, and NHPs. NHPs are attractive due to their high degree of genomic and physiologic similarity to humans, including natural inter-individual genetic variation. Previous reports have proven that rhesus macaques serve as durable and clinically relevant animal model to study infectious diseases and cancers [9,10,17,18]. In this study, our results from MSI testing, IHC, gene expression profiling, systems biology approaches, somatic variant calling, and DNA methylation of colon tissue samples from the KCCMR cohort demonstrated that rhesus macaques develop CRC phenotypes analogous to MSI, including both sporadic MSI-H and LS patients. These findings indicate that rhesus macaques may serve as a useful animal model for studying MMRd CRC and address some of the shortcomings of previously established model systems mostly linked to immune-host interactions due to emergence of somatic mutations from MMR deficiency. Nonetheless, it is essential to acknowledge that rhesus, and NHPs in general, have limitations and disadvantages including barriers of high cost, availability, ethical concerns, housing requirements, longer carcinogenic intervals, and intrinsic biological differences.

To characterize the rhesus macaque as a surrogate for studies of MMRd, we investigated the MSI status of 6 markers across 41 unique rhesus tumors using a newly designed, in-house MSI panel for rhesus CRC. Our study results indicated that 76% of rhesus CRC from the KCCMR cohort displayed an MSI-H phenotype, which warrants the use of rhesus as an optimal system to study MMRd carcinogenesis. Many rhesus tumors lost expression of MLH1 and PMS2 proteins, but retained the expression of MSH2 and MSH6, as confirmed by IHC analysis. The MLH1 germline stop codon mutation (c.1029C>G, p.Tyr343Ter), previously reported as a pathogenic variant in human LS (National Center for Biotechnology Information), was present in 8 (19.5%) rhesus macaques, while the majority (80.5%) were wild-type for this variant. The majority of CRC from MLH1 mutation carriers presented an MSI-H phenotype; however, two tumors displayed an MSI-L status. This finding is consistent with previous observations in LS patients and reflects that LS carcinogenesis can follow the canonical MMRd route, or a MMRd pathway that is more frequently observed in sporadic carcinogenesis driven by WNT activation [19].

DNA methylation analysis of rhesus CRC suggested that epigenetics plays a pivotal role in the rhesus CRC development. Comparative DNA methylation from colon tumor and adjacent normal tissue samples indicated clear segregation of methylation patterns between MSI-H and MSI-L/MSS CRC samples and also between tumor and adjacent normal mucosa. Interestingly, although human CRC typically displays widespread DNA methylation through the promoter region of the MLH1 gene, methylation of rhesus CRC predominantly occurred in the exon 1 of MLH1.

Despite prior reports of tissue-specific transcriptome analysis of fresh frozen tissues from rhesus macaques, no study had previously profiled the colonic tissue from rhesus macaques of Indian origin [20]. Therefore, this constitutes the first analysis of matched tumor and normal colon samples in rhesus macaques using NGS. Our analysis observed clear expression differences between rhesus tumor and normal samples, and when compared to human CRC data from TCGA, rhesus MSI-H tumors were more similar to human MSI-H than MSS tumors. These findings of transcriptomic similarity between humans and rhesus CRC support utilizing data derived from rhesus LS to study aspects of MMRd carcinogenesis that requires assessment of global transcription patterns such as neoantigen discovery and profiling of CRC. Moreover, to confirm the biological relevance of the rhesus macaque as an animal model, we performed CMS classification and GSEA to ascertain the molecular features of rhesus MSI and LS CRCs. Rhesus CRC mainly associated with CMS2, which is the canonical CRC subtype that corresponds with high levels of copy number changes and activation of WNT/MYC pathways [15]. However, rhesus LS tumors primarily associated with CMS1 (MSI-Immune), which encompasses MSI, CpG Island Methylator Phenotype (CIMP) high, hypermutation, and immune activation, thus aligning with previous studies from our group [21]. Conversely, most human sporadic CRCs typically cluster with CMS2, which was also observed in a relevant fraction of sporadic rhesus tumor samples. We acknowledge that this observation is not entirely consistent with results from human CRC, but it could reflect that the CMS classifier has been optimized to classify human tumors and it would require computational adaptation to rhesus data to consider intrinsic differences in rhesus tumorigenesis, which is predominantly metastatic. This advanced stage of tumorigenesis at time of detection in rhesus may explain the predominance of CMS2-associated signals, at which point an early-stage CMS1 tumor may have converged toward CMS2 classification due to late-staged WNT activation.

GSEA indicated activation of key pathways—namely cancer stem cell (CSC) signatures and crypt base—in sporadic MSI rhesus CRC, which corroborates a previously described signature of human MMRd CRC [22]. The pathway enrichment between MSI-L/MSS and MSI-H indicates that these advanced, late-stage lesions are transcriptomically similar, which may be driven by the late time point rather than MSI status. These findings provide evidence to support the use of rhesus macaques as a model to understand the molecular basis and tumor micro-environment in MMRd tumorigenesis.

To quantify the mutational rate in rhesus MMRd CRCs, we leveraged RNAseq data of the rhesus LS tissues. This allowed us to observe high mutation rates in genes commonly mutated in CRC, thus adding additional support to the case for utilization of rhesus macaques for vaccine research and immunotherapy development, since there is strong expression of tumor-associated neoantigens derived from observed somatic mutations. In addition, the high average tumor purity in RNAseq (>65%) along with the mean variant allele frequency (VAF) increased the power of our mutation analysis and provided assurance on the level of detection somatic mutations rhesus CRC from RNAseq.

We report a spontaneous NHP model for MLH1-mutated Lynch Syndrome and more generally sporadic CRC. An important point to consider in this model is the ability of NHP colonies to maintain a high level of environmental uniformity and patterned genetic relatedness between pedigreed individuals, which affords a deeper understanding of the contributions of genes in complex disease that are not comparably possible with other genetically engineered models or even human populations. Although, gene editing technologies such as CRISPR/Cas9 seem to have enormous potential to develop novel biomedical research models, NHP models have not been widely used for such gene editing technologies in biomedical research at the present time. In contemplating the future possibility, feasibility, and value of creating targeted gene editing of MLH1, MSH2, MSH6, PMS, or EPCAM in a NHP with the intent to model MMRd human LS, one should consider the expected length of time for onset for the disease phenotype. Other considerations should include the relative penetrance seen in human LS with each of the MMR genes. MSH6 and PMS2 lead to phenotypes that are less penetrant in humans. MSH2 is as penetrant as MLH1 in humans, but more frequently leads to ovarian or endometrial cancer. Such considerations should be balanced a priori against what questions would be expected to be answered with a gene-edited NHP that could not be potentially answered by a spontaneous model.

We acknowledge that our study has several limitations necessitating further investigation. Importantly, the comparator group, MMR proficient (MMRp) tumors, only included one animal, which challenged the validity of the comparison between MMR proficiency and deficiency. Thus, a stronger comparator group is necessary to strengthen our findings. Furthermore, this study lacks pertinent information regarding the timeline of carcinogenesis for both sporadic and LS rhesus tumors, which restricts our understanding of pre-cancer biology, and the timing of tumor development and subsequent evolution. In addition, neoantigen detection and T-cell receptor (TCR) profiling would be an important asset for a complete understanding of the immune system in the rhesus macaque CRC. Lastly, our mutation calling was performed using total RNA sequencing data, which, although adequate, is less ideal than whole exome sequencing.

In conclusion, this study provides a robust molecular and genetic characterization of a spontaneous and translationally relevant NHP animal model that will be useful for understanding MMRd CRC, including LS CRC. These results justify the preclinical use of the rhesus to study LS CRC and also the larger group of sporadic MSI CRCs in specific contexts, such as survey the immune landscape, discovery of prevention strategies, assessment of TME dynamics, development of treatment towards advanced cancers, all in a model system with moderate to high translational value to humans.

[END]

[1] Url: https://journals.plos.org/plosgenetics/article?id=10.1371/journal.pgen.1010163

(C) Plos One. "Accelerating the publication of peer-reviewed science."
Licensed under Creative Commons Attribution (CC BY 4.0)
URL: https://creativecommons.org/licenses/by/4.0/


via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/plosone/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.